AI & ML interests

Smart monitoring and control systems that grow under administrator approval, extended by a robot simulation studio with self QA/QC and on-premise small LLMs.

Recent Activity

yuncheol-ncdtechย  updated a Space about 5 hours ago
NCDTech/README
yuncheol-ncdtechย  updated a model about 6 hours ago
NCDTech/sim-driving-mlp-numpy
yuncheol-ncdtechย  published a model about 6 hours ago
NCDTech/sim-driving-mlp-numpy
View all activity

Organization Card

NCDTech โ€” We design and research smart monitoring and control systems that grow.

The NCDTech website

(Our website and the diagrams and screens below are in Korean; they are the same figures used on our website and demo video.)

Our roots are in smart monitoring and control.

With our robot simulation studio we stand on that extension line, and our AI sits on top of it.

NCDTech 4-layer stack

Conventional monitoring and control ends at collecting data, showing it, and raising alarms on preset conditions.

Interpretation and judgment are left to the experts who know the system.

What we build is the layer above that: self-inspection wired into every stage, and an on-premise LLM that reads those inspection records and explains them in plain conversation.

An operator conversation with the on-premise LLM, captured as a knowledge record

Making code verify itself is nothing new.

Test benches and unit tests live outside the code; self-checks live inside it: assertions, Design by Contract, and built-in self-tests (BIST/POST) are standard embedded practice.

We have been putting such self-check devices into code for 24 years of embedded measurement and control work.

Our technology is the wiring above that standard: the check records feed an on-premise LLM, so the system reports in plain language, holds a conversation, and grows its knowledge under administrator approval.

The system grows through use, but only under administrator approval: code-based verdicts are never overwritten, every approved piece of knowledge keeps its full revision history, and nothing is ever silently deleted.

The administrator approval gate: accept, reject, or revise as a new revision

We do not so much invent new capabilities as put AI, without hesitation, on top of abilities we have already verified, and make it actually run.

Current status: first development phase complete, applied to our in-house autonomous driving simulator.

Demo application in progress, with the knowledge base still growing.

The in-house simulator: a tiny neural network driving in the 3D evaluation stage

The tracked test robot the simulator reproduces, with hand-mounted ultrasonic sensors

The same robot driving on the test floor, monitored live by our control dashboard


ํ•œ๊ตญ์–ด

NCDTech โ€” ์„ฑ์žฅํ•˜๋Š” ์Šค๋งˆํŠธ ๋ชจ๋‹ˆํ„ฐ๋ง ์ œ์–ด ์‹œ์Šคํ…œ์„ ์„ค๊ณ„ํ•˜๊ณ  ์—ฐ๊ตฌํ•˜๋Š” ํšŒ์‚ฌ์ž…๋‹ˆ๋‹ค.

์ €ํฌ์˜ ๋ฟŒ๋ฆฌ๋Š” ์Šค๋งˆํŠธ ๋ชจ๋‹ˆํ„ฐ๋ง๊ณผ ์ œ์–ด์ž…๋‹ˆ๋‹ค.

๋กœ๋ด‡ ์‹œ๋ฎฌ๋ ˆ์ด์…˜ ์ŠคํŠœ๋””์˜ค๋กœ ๊ทธ ํ™•์žฅ์„  ์œ„์— ์žˆ๊ณ , AI๋Š” ๊ทธ ์œ„์— ์˜ฌ๋ผ๊ฐ‘๋‹ˆ๋‹ค.

๊ธฐ์กด์˜ ๋ชจ๋‹ˆํ„ฐ๋งยท์ œ์–ด๋Š” ๋ฐ์ดํ„ฐ๋ฅผ ๋ชจ์•„ ๋ณด์—ฌ์ฃผ๊ณ  ์ •ํ•ด์ง„ ์กฐ๊ฑด์œผ๋กœ ๊ฒฝ๋ณด๋ฅผ ๋‚ด๋Š” ๋ฐ๊นŒ์ง€์ด๊ณ , ํ•ด์„๊ณผ ํŒ๋‹จ์€ ์‹œ์Šคํ…œ์„ ์•„๋Š” ์ „๋ฌธ๊ฐ€์˜ ๋ชซ์œผ๋กœ ๋‚จ์Šต๋‹ˆ๋‹ค.

์ €ํฌ๊ฐ€ ๊ตฌํ˜„ํ•˜๋Š” ๊ฒƒ์€ ๊ทธ ์œ„ ๋‹จ๊ณ„์ž…๋‹ˆ๋‹ค: ๋ชจ๋“  ๋‹จ๊ณ„์— ์‹ฌ์€ ์ž๊ฐ€๊ฒ€์‚ฌ, ๊ทธ๋ฆฌ๊ณ  ๊ทธ ๊ฒ€์‚ฌ ๊ธฐ๋ก์„ ์ฝ๊ณ  ๋Œ€ํ™”๋กœ ์„ค๋ช…ํ•˜๋Š” ์˜จํ”„๋ ˆ๋ฏธ์Šค LLM ์ž…๋‹ˆ๋‹ค.

์ฝ”๋“œ๊ฐ€ ์Šค์Šค๋กœ๋ฅผ ๊ฒ€์ฆํ•˜๊ฒŒ ๋งŒ๋“œ๋Š” ์ผ์€ ์ƒˆ๋กœ์šด ๊ฒƒ์ด ์•„๋‹™๋‹ˆ๋‹ค.

ํ…Œ์ŠคํŠธ ๋ฒค์น˜ยท์œ ๋‹› ํ…Œ์ŠคํŠธ๋Š” ์ฝ”๋“œ ๋ฐ–์— ์žˆ๊ณ , ์ž๊ฐ€๊ฒ€์‚ฌ๋Š” ์ฝ”๋“œ ์•ˆ์— ์žˆ์Šต๋‹ˆ๋‹ค: assertionยทDesign by ContractยทBIST/POST ๊ฐ™์€ ์ž„๋ฒ ๋””๋“œ ํ‘œ์ค€ ๊ธฐ์ˆ ์ž…๋‹ˆ๋‹ค.

์ €ํฌ๋Š” 24๋…„์˜ ์ž„๋ฒ ๋””๋“œ ๊ณ„์ธกยท์ œ์–ด ์ผ์—์„œ ๊ทธ๋Ÿฐ ์ž๊ธฐ ์ ๊ฒ€ ์žฅ์น˜๋ฅผ ์ฝ”๋“œ ์•ˆ์— ๋„ฃ์–ด ์™”์Šต๋‹ˆ๋‹ค.

์ €ํฌ ๊ธฐ์ˆ ์€ ๊ทธ ํ‘œ์ค€ ์œ„์˜ ์—ฐ๋™์ž…๋‹ˆ๋‹ค: ๊ฒ€์‚ฌ ๊ธฐ๋ก์„ ์˜จํ”„๋ ˆ๋ฏธ์Šค LLM์— ์—ฐ๊ฒฐํ•ด, ์‹œ์Šคํ…œ์ด ์‚ฌ๋žŒ ๋ง๋กœ ๋ณด๊ณ ํ•˜๊ณ  ๋Œ€ํ™”ํ•˜๋ฉฐ ๊ด€๋ฆฌ์ž ์Šน์ธ ์•„๋ž˜ ์ง€์‹์ด ์„ฑ์žฅํ•˜๊ฒŒ ๋งŒ๋“œ๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.

์‹œ์Šคํ…œ์€ ์“ธ์ˆ˜๋ก ์„ฑ์žฅํ•˜๋˜, ๋ฐ˜๋“œ์‹œ ๊ด€๋ฆฌ์ž ์Šน์ธ ์•„๋ž˜์—์„œ๋งŒ ์„ฑ์žฅํ•ฉ๋‹ˆ๋‹ค: ์ฝ”๋“œ ๊ธฐ๋ฐ˜ ํŒ์ •์€ ์ ˆ๋Œ€ ๋ฎ์–ด์“ฐ์ง€ ์•Š๊ณ , ์Šน์ธ๋œ ์ง€์‹์€ ์ „์ฒด ๊ฐœ์ • ์ด๋ ฅ์„ ๋ณด์กดํ•˜๋ฉฐ, ์–ด๋–ค ๊ธฐ๋ก๋„ ์กฐ์šฉํžˆ ์‚ญ์ œ๋˜์ง€ ์•Š์Šต๋‹ˆ๋‹ค.

์—†๋˜ ๊ธฐ๋Šฅ์„ ์ƒˆ๋กœ ๋ฐœ๋ช…ํ–ˆ๋‹ค๊ธฐ๋ณด๋‹ค, ๊ฒ€์ฆ๋œ ๋Šฅ๋ ฅ ์œ„์— AI๋ฅผ ๊ฑฐ๋ถ€๊ฐ ์—†์ด ์–น์–ด ์‹ค์ œ๋กœ ๋™์ž‘์‹œํ‚ค๋Š” ๊ฒƒ์ž…๋‹ˆ๋‹ค.

ํ˜„์žฌ ์ƒํƒœ: 1์ฐจ ๊ฐœ๋ฐœ ์™„๋ฃŒ, ์ž์‚ฌ ์ž์œจ์ฃผํ–‰ ์‹œ๋ฎฌ๋ ˆ์ดํ„ฐ ์ ์šฉ.

๋ฐ๋ชจ ์ ์šฉ ์ค‘์ด๋ฉฐ ์ง€์‹์€ ํ™•์žฅ ์ค‘์ž…๋‹ˆ๋‹ค.


Website: https://www.ncdtech.org

LinkedIn: https://www.linkedin.com/company/112725964

Dataset: NCDTech/human-gated-qaqc-knowledge-example ยท Model: NCDTech/sim-driving-mlp-numpy